Arm Action Planning and Pick-and-Place Objects for Humanoid Robot

Abstract

In the future, when the labor force is reduced and costs are rising, a humanoid robot can be used to perform work that was originally for a human. In this paper, the function of object gripping is implemented on a small humanoid robot platform Image recognition is used to determine the position of objects on the table and the objects are clamped to the target area in accordance with the planned sequence. First, the robot stands in front of the table and uses the camera to identify the position of the object on the table. Then, it controls the arm to grasp the object and finally places the object in the collection area. Table tennis balls and paper cups are used to represent two different types of garbage. The robot judges the objects and order of picking, and places the objects in a trash can of the respective color next to the table. The computing performance of the robot is not very good, so this paper combines the greedy heuristic method and the traveling salesman method to plan the clamping sequence of the shortest distance. This method also reduces collisions of the arm with other objects on the table when the grasped object moves, avoids the object being overturned or dropped on the floor, and successfully achieves the task of grasping different objects and classifying them.